Transitions from Drag-based to Lift-based Propulsion in ...

AMER. ZOOL.,36:628-641 (1996)

Transitions from Drag-based to Lift-based Propulsion in
Mammalian Swimming'

FRANK E. FISH

Department of Biology, West Chester University,
West Chester, Pennsylvania 19383

SYNOPSIS. The evolution of fully aquatic mammals from quadrupedal,
terrestrial mammals was associated with changes in morphology and
swimming mode. Drag is minimized by streamlining body shape and
appendages. Improvement in speed, thrust production and efficiency is
accomplished by a change of swimming mode. Terrestrial and semiaquatic
mammals employ drag-based propulsion with paddling appendages,
whereas fully aquatic mammals use lift-based propulsion with oscillating
hydrofoils. Aerobic efficiencies are low for drag-based swimming, but
reach a maximum of 30% for lift-based propulsion. Propulsive efficiency
is over 80% for lift-based swimming while only 33% for paddling. In
addition to swimming mode, the transition to high performance propul-
sion was associated with a shift from surface to submerged swimming
providing a reduction in transport costs. The evolution of aquatic mam-
mals from terrestrial ancestors required increased swimming performance
with minimal compromise to terrestrial movement. Examination of mod-
ern analogs to transitional swimming stages suggests that only slight mod-
ification to the neuromotor pattern used for terrestrial locomotion is re-
quired to allow for a change to lift-based propulsion.

INTRODUCTION aquatic mammals and their transitional
forms, functional analysis is therefore nec-
Numerous paleontological, morphologi- essary to determine the possible environ-
cal, and molecular studies have focused on mental factors that produced the various
the phylogenetic relationships both within aquatic adaptations.
clades of aquatic mammals and with their
terrestrial ancestors (Barnes et al., 1985; The similarity in morphology and swim-
Wyss, 1988, 1989; Berta, 1991; Milinkov- ming mode between fish and cetaceans is
itch et al., 1993; Adachi and Hasegawa, the quintessential example of evolutionary
1995). Such studies were helpful in pro- convergence (Howell, 1930; Fish, 1 9 9 3 ~ ) .
ducing phylogenies and in elucidating the This familiar textbook example illustrates
historical sequence of structural changes. how similar functional requirements are
However, phylogenetic studies are limited met by organisms that do not share a single
and lack a functional approach to determine clade. Such convergence, with its resulting
the causal reasons for evolutionary change homoplasy, is associated with similar con-
(Lauder, 1990). Functional analysis pro- straints imposed on animals by the physical
vides an understanding at a mechanistic environment and with selection for adapta-
level through measures of performance tions for effective swimming (Fish, 1 9 9 3 ~ ) .
(Le., exercise metabolism, locomotor kine-
matics, maximal swimming speed). For Adaptations associated with increased
aquatic behavior and swimming perfor-
I From the Symposium Aquatic Locomotion: New mance evolved independently in several
Approaches ro lnverrebrare and Vertebrate Biome- clades of mammals (Le., Cetacea, Pinnipe-
chanics presented at the Annual Meeting of the Society dia, Sirenia). Modem groups of aquatic
for Integrative and Comparative Biology, 27-30 De- mammals display swimming adaptations
cember 1995, at Washington, D.C. that optimize use of energy by reduction in
drag and improvement in thrust production

628

I.

FROM DRAG-BASED TO LIFT-BASED PROPULSION 629

and efficiency (Lang, 1966; Williams, ratio (FR), where FR is body lengthlmaxi-

1989; Fish, 1993a, b). Drag is minimized mum thickness. The optimal FR, which

by streamlining the body and appendages provides the minimum drag for the maxi-

(Webb, 1975; Fish, 1 9 9 3 ~ )T. hrust and ef- mum volume, is 4.5 (von Mises, 1945).

ficiency are maximized by swimming However, FR can vary from 3 to 7 with

modes that use a lift-based oscillating hy- only a 10% increase in drag (Webb, 1975).
drofoil (i.e., cetacean flukes, phocid hind- Use of FR as an index of locomotor per-

flippers, otariid foreflippers). Alternatively, formance is limited. Mammals have FRs
less aquatically adapted terrestrial and within the range of reduced drag regardless
semiaquatic mammals employ drag-based of aquatic ability (Fish, 1 9 9 3 ~ ) .Sirenians
paddling (Howell, 1930; Williams, 1983; have FRs which span the optimal value of
Fish, 1992, 1 9 9 3 ~ )t;his form of propulsion 4.5. Although able to sprint at speeds of 6
d s e c , these herbivores swim slowly at 0.6-
is considered of lower performance and 0.8 d s e c (Hartman, 1979; Nishiwaki and
more primitive than the lift-based mode Marsh, 1985). Fish (1993a) suggested that
(Tarasoff et al., 1972; Baudinette and Gill, FR for sirenians may be associated with a
1985; Williams, 1989; Fish, 1 9 9 3 ~ ) . body design to limit heat loss by a ther-
mally sensitive aquatic species with low
The evolution of highly derived aquatic metabolism. High values of FR in mustelids
morphologies and lift-based swimming may be more an indication of a body orig-
modes represents the culmination of a se- inally designed for hunting in burrows than
quence of transitional stages from terrestrial for swimming (Williams, 1983, 1989). The
quadrupeds to fully aquatic mammals greatest range of FR (4-1 1) is found in the
(Howell, 1930; Barnes et al., 1985; Gin- cetacean family Delphinidae from Dall’s
gerich et al., 1990, 1994; Fish, 1992). The porpoise (Phocoenoides dalli) to the north-
intermediates between fully aquatic mam- ern right whale dolphin (Lissodelphis bo-
mals and their terrestrial ancestors were be- realis) (Fish, 1 9 9 3 ~ ) .Despite the diver-
lieved to be semiaquatic that swam by pad- gence in FR,these two species are consid-
dling (Barnes et al., 1985; Thewissen et al., ered among the fastest dolphins with max-
1994). In that the kinematics and mechanics imum speeds exceeding 8 d s e c (Winn and
of thrust generation differs between drag- Olla, 1979).
based and lift-based propulsion, how was
the transition in swimming mode achieved A better indicator of swimming perfor-
in the evolution of aquatic mammals? In an- mance based on design is streamlining by
swering this question, clues may be eluci- the possession of a fusiform body. The fu-
dated as to what selection forces and inter- siform shape is characterized by a rounded
mediate morphological stages were re- leading edge with a long tapering tail.
quired in the transition toward increased Streamlining minimizes drag by reducing
aquatic behavior. the magnitude of the pressure gradient over

BODY DESIGN the body allowing water to flow over the
surface without separation (Vogel, 1994).
Body design has been one of the princi- Unfortunately, the extent of streamlining in
ple morphological parameters examined as mammals is not easily measured. Profiles of
an index of adaptation to the aquatic envi- dolphins and sea lions were compared with
ronment. Body shape is associated with the two-dimensional airfoils (Hertel, 1966;
amount of force required to overcome drag Feldkamp, 1 9 8 7 ~ )H. owever, such matches
(resistive force); a factor influenced by the only reflect a crude resemblance and are not
flow pattern around the body. Since drag is exact due to the three-dimensional config-
equal to thrust (propulsive force) when the uration of the animal and its deviation from
body is moving at constant velocity, body
shape has been used to characterize the the smooth fusiform shape.

swimming performance of animals. SWIMMING MODE MECHANICS AND
The most general index of body design DIVERSITY

used to estimate hydrodynamic perfor- Aside from body design, swimming

mance (i.e., swimming speed) is fineness modes determine how effectively an animal

I

630 FRANK E. FISH

TABLE 1. Swimming modes exhibited by modern mammals.

Mode Thrust force Representative genera Habits

Quadrupedal paddling Drag Didelphis, Canis, Elephas, terrestrial, semiaquatic
Mustela
Alternate pectoral paddling Drag semiaquatic
Alternate pelvic paddling Drag Mustela, Thalarctos semiaquatic
Castor, Chironectes,
Alternate pelvic rowing Drag semiaquatic
Simultaneous pelvic paddling Drag Hydromys, Ondatra semiaquatic
Pectoral rowing Drag Ondatra semiaquatic
Lateral undulation Acceleration Lutra semiaquatic
Ornithorynchus
Dorsoventral undulation reaction Potomogale semiaquatic, aquatic
Acceleration
Pelvic oscillation Enhydra, Lutra aquatic
Pectoral oscillation reaction aquatic
Caudal oscillation Lift Odobenus, Phoca aquatic
Lift Zalophus
Lift Trichechus, Tursiops

moves through water (Fish, 1 9 9 3 ~ )T. wo rected perpendicular to the pathway tra-
distinct swimming modes are exhibited by versed by the hydrofoil and can be resolved
mammals, drag-based and lift-based pro- into an anteriorly directed thrust force
pulsion. Drag-based swimming is consid- (Weihs and Webb, 1983; Fish, 1993~).
ered a primitive, low performance mode, Thrust is generated continuously through-
whereas lift-based propulsion is a high per- out a stroke cycle. Although some drag is
formance, derived mode based on undula- produced by the hydrofoil, it is small com-
tion of the body (Howell, 1930; Fish, pared to the lift. The high lift-to-drag ratio
1993~). is a function of the high aspect ratio (span2/
planar area) of the hydrofoil (Feldkamp,
Drag-based propulsion is used by a va- 1 9 8 7 ~F; ish et al., 1988; Fish, 1993a, b).
riety of terrestrial and semiaquatic mam-
mals (Howell, 1930; Fish, 1 9 9 3 ~ ) .The Cetaceans and sirenians are caudal oscil-
limbs are oriented in either the vertical par- lators (Fish, 1 9 9 3 ~ )T. his mode also is de-
asagittal plane (paddling) or the horizontal scribed as thunniform or carangiform with
plane (rowing), and strokes of the paired lunate tail (Lighthill, 1969; Webb, 1975;
appendages are either alternate or simulta- Lindsey, 1978) similar to certain fast swim-
neous (Table 1). Stroke cycle is divided into ming fish. Unlike fish, the mammals move
power and recovery phases (Fish, 1984). their propulsive flukes in the sagittal plane
During the power phase, the posterior (Hartman, 1979; Nishiwaki and Marsh,
sweep of the foot generates drag which is 1985; Fish and Hui, 1991). Angle of attack
used to provide an anterior thrust for the is controlled by a joint at the base of the
animal (Fig. 1). Paddle area is increased by flukes to maximize thrust throughout the
abduction of the digits and interdigital web- stroke. The Otariidae, including sea lions
bing or fringe hairs (Howell, 1930; Fish, and fur seals, use foreflippers as oscillatory
1984, 1 9 9 3 ~ ) .The recovery phase reposi- hydrofoils (English, 1976; Feldkamp,
tions the foot and incurs a non-thrust gen- 1987a, b). The stroke is described as pec-
erating drag. To prevent negating the thrust toral oscillation (Fish, 1 9 9 3 ~ a) nd as sub-
during the recovery phase, drag on the foot aqueous flight similar to swimming by pen-
is reduced by adducting the digits to reduce guins and sea turtles (Baudinette and Gill,
paddle area (Fish, 1984). 1985; Feldkamp, 198721). Phocid seals and
the walrus use pelvic oscillation for swim-
Highly derived aquatic mammals pro- ming. The posterior body is laterally un-
duce lift to generate thrust (Fig. 1). Lift is dulated moving the paired hindflippers in
generated by oscillating a hydrofoil (Le., the horizontal plane (Backhouse, 1961;
flukes, flippers) at a controlled angle of at- Ray, 1963; Tarasoff et al., 1972; Gordon,
tack (Lighthill, 1969; Feldkamp, 1987b; 1981; Fish et al., 1988). Ankle joints con-
Fish et al., 1988; Fish, 1993b). Lift is di-

II

FROM DRAG-BASED TO LIFT-BASED PROPULSION 63 1

undulating a deeply compressed tail (Walk-
er, 1975). Lateral undulations of the com-
pressed tail of the muskrat (Orzdatra) pro-
vided a maximum of 2% of the thrust,
which was produced mainly by paddling of
the hindfeet (Fish, 1 9 8 2 ~ ) .Otters are re-

L ported to propel themselves by dorsoven-
7 trally undulating the body and tail (Kenyon,

1969; Tarasoff et al., 1972; Williams,
1989). Fish (1994) demonstrated in the riv-
er otter, Lutra canadensis, that simulta-
neous paddling by the hindfeet is integral
to the undulatory mode.

L SWIMMING MODE ENERGETIC PERFORMANCE
d
If aquatic mammals are adapted to swim
1 in a manner that minimizes energy expen-
diture, there should be distinct metabolic
FIG. 1. Major forces associated with propulsive and hydrodynamic advantages to propulsive
modes. From the top, animals and propulsive modes modes employed by the most derived spe-
shown are (1) muskrat (Ondarra zibethicus):paddling, cies. To understand the performance asso-
(2) sea lion (Zaluphus californianus): pectoral oscil- ciated with increasing aquatic habits, the
lation, (3) harp seal (Phuca groenlandica): pelvic os- energetics of swimming can be examined
cillation, and (4) bottlenose dolphin (Tursiops trun- by comparing efficiencies with regard to
catus): caudal oscillation. To swim at a constant ve- propulsive modes and morphologies of a
locity, an animal must generate a thrust force (T) equal wide variety of mammals. Maximization of
and opposite to a drag force (D). Paddling is a drag- thrust production and increased efficiency
based swimming mode. The posterior sweep of the are highly associated with swimming
hind feet of the muskrat generates a drag force (d) modes (Rayner, 1985; Fish, 1992).
which contributes to T. Using lift-based propulsive
modes, the foreflippers of the sea lion, hind flippers of Efficiency can relate active metabolism
the harp seal, and caudal flukes of the dolphin act as with total mechanical work and propulsive
hydrofoils to generate T. Lift derived from movement power output (Fish, 1993a). There is a pau-
of the hydrofoil is resolved into an anteriorly directed city of data on efficiency because studies of
thrust force. The drag associated with the hydrofoil (d) swimming performance rarely examine
is produced mainly from frictional and induced drag both metabolism and mechanical work.
components and is small relative to the lift force. Aerobic efficiency (qa) is power output
(From Fish, 1993a.) (thrust power) divided by aerobically sup-
plied power input (active metabolic rate);
trol the angle of attack, analogous to ceta- that is the ratio of power expended in doing
ceans and thunniform fish. useful work (Le., move the animal forward
through the water) to potential power avail-
Intermediate between drag-based and able to perform work. Propulsive efficiency
lift-based propulsion is the undulatory (qp) is the ratio of thrust power to rate of
mode (Webb, 1988). The African otter total mechanical work produced. Drag pow-
shrew (Potamogale) may swim by laterally er may be substituted for the thrust power
when the animal is swimming steadily.
However, use of drag values may give an
underestimate of efficiency, because pro-
pulsive movements can increase total drag
on the animal.

Comparison of efficiencies between dif-
ferent aquatic species show that lift-based
modes are higher than drag-based paddling

632 FRANK E. FISH

.c

w

.0-

P

Species

FIG. 2. Comparison of efficiencies between drag-based (black bars) and lift-based (hatched bars) swimming
modes. Aerobic efficiencies are displayed in A; propulsive efficiencies in B. Data from DiPrampero et al., (1974).
Webb (1975), Williams (1983, 1989); Fish (1984, 1985, 1992, 1993b, unpublished data), Feldkamp (1987a),
Fish er a!., (1988), Bose and Lien (1989). Williams et al., (1992).

(Fig. 2). Values of qafor paddling mammals dulatory swimming in mammals, but esti-
are no better than 5%, whereas lift-based mates of qpfor trout by Webb (1978) give
swimming typically reach 20% with a max- an intermediate value of 67%.
imum value for sea lions of 30%. A similar
trend is exhibited for qp (Fig. 2). Paddlers Cost of Transport (CT) can be used to
had maximum reported values for qp of assess the efficiency of different swimming
33%. The -qp of lift-based swimmers is at modes independent of power output
least 80%. No data are available for the un- (Schmidt-Nielsen, 1972; Williams, 1989;
Fish, 1992). CT is the metabolic cost to

FROM DRAG-BASED TO LIFT-BASED PROPULSION 633

10 : A 0 Human

0 44 A Mink
0 Muskrat
4
4 Sea otter (surface)

FIG. 3. Relationship between Cost of Transport (CT) and body mass. Open symbols represent CT values for
drag-based propulsion when surface swimming; closed symbols represent undulatory and lift-based propulsion
during submerged swimming. The solid line represents the extrapolated minimum CT for fish (Davis et al.,
1985). Data from Costello and Whittow (1975); Kruse (1975); 0ritsland and Ronald (1975); E! E. DiPrampero,
personal communication (1979); Fish (1982); Sumich (1983); Williams (1983); Innes (1984); Davis et al. (1985);
Worthy et al. (1987); Feldkamp (1987a); Williams (1989); Williams et al. (1992).

move a unit mass a given distance and CT tile Varanus and mammal Didelphis
is inversely proportional to efficiency showed similar activity patterns between
(Tucker, 1970). Drag-based swimmers have homologous muscles with the same attach-
higher values of CT compared to lift-based ment sites (Jenkins and Goslow, 1983). Fur-
swimmers of similar body mass (Fig. 3). thermore, homologous muscles with differ-
Compared to the extrapolated line for min- ent attachments had similar activity pat-
imum CT of fish (Davis et al., 1985), pad- terns. Neuromotor conservation was dem-
dling mammals have values of CT 10-25 onstrated also between walking and flight
times greater than fish of similar mass; (Goslow et al.,1989; Dial et al.,1991). Al-
whereas lift-based swimmers are only 1.9- though muscles have been functionally re-
4.6 times greater. A particularly significant organized for flight, several muscles have
data set was collected by Williams (1989) similar timing patterns in their respective
on sea otters showing the energetic advan- locomotor cycle with walking gaits (Dial et
tage of shifting from paddling at the water
surface to submerged swimming using un- al., 1991).
dulation. CT of surface paddling was 69% The basis for conservation of neuromotor
greater than submerged swimming.
pattern is the existence of central pattern
ASSOCIATION BETWEEN TERRESTRIAL AND generators. Central pattern generators are
AQUATIC MOVEMENTS neural networks in the central nervous sys-
tem that govern specific, repetitive motions
Neuromotor patterns for locomotion ap- by providing the correct timing for mus-
pear to be conservative (Jenkins and Gos- cular contraction (Grillner, 1975, 1996; Pe-
low, 1983; Goslow et al., 1989; Smith, ters, 1983; Pearson, 1993). Motor patterns
1994) and are independent of changes in are executed without peripheral feedback,
morphology and functional role (Smith, although afferent feedback can modify the
1994). Kinematics and electromyographic motor programming and its output (Pear-
analysis of terrestrial locomotion in the rep- son, 1993).

The rhythmic nature of the central pat-

634 FRANK E. FISH

Terrestrial Aquatic Propulsive Selection
Parameters
Motor Pattern Mode

Asymmetrical Gaits *Lift-Based
(Gallop, Bound) Oscillation
t
Speed
Power
Efficiency

Symmetrical Gaits Drag-Based
(Walk, Trot) Paddling

FIG. 4. Association between terrestrial and aquatic propulsive mode in relation to selection parameters for
enhance performance in swimming.

tern generator is considered important in Lift-based oscillatory modes arise pri-
the automatic movements associated with marily from asymmetrical terrestrial gaits
locomotion (Grillner and Wallen, 1985; (Fig. 4). Such land-based gaits may be the
Pearson, 1993; Stein et al., 1995). Central most economical high-speed gaits, because
pattern generators have been demonstrated of potential recycling of elastic strain en-
for walking and running in mammals and ergy (Alexander, 1988). Energy economy
birds, flight in birds, and swimming in fish by storage and recovery of elastic energy in
and reptiles. The diversity of vertebrates a flexing body has been the focus of some
and locomotor behaviors linked with central work on dolphin swimming mechanics
pattern generators alludes to the generality (Bennett et al., 1987; Blickhan and Cheng,
of these neural networks. 1994; Pabst, 1996). The undulatory mode
begins as a bound or half bound using pow-
Despite dissimilarities in kinematics and erful simultaneous strokes of the forelimbs
performance between drag-based and lift- and hindlimbs enhanced by spinal flexion
based propulsion, these modes can be (Fish, 1994). After an initial acceleration,
linked through the conservation of primi- only hindlimbs were used. This mode is as-
tive neuromotor patterns associated with sociated with rapid swimming matching the
terrestrial gaits (Fig. 4). Symmetrical ter- use of asymmetrical gaits at high speeds on
restrial gaits such as walks are the basis for land. The sea otter, Enhydra lutris, uses spi-
drag-based swimming in mammals. Exam- nal flexion in concert with simultaneous
ination of swimming by semiaquatic and strokes of the hindfeet (Kenyon, 1969; Tar-
terrestrial mammals show that they quad- asoff et al., 1968; Williams, 1989), but the
rupedally paddle in gaits reminiscent of a short tail precludes thrust generation by this
diagonal sequence run, trot, and lateral se- structure, necessitating enlargement and use
quence singlefoot (Fig. 5; Williams, 1983; of the hindlimbs as propulsors. For river ot-
Fish, 1 9 9 3 ~ )T. he “dog paddle” is actually
ters, Lutra canadenesis, simultaneous
a modified lateral sequence run (personal strokes of the hindlimbs generate a thrust-
observation). In humans, the crawl stroke producing travelling wave in the body and
with a two-beat kick is similar to a diagonal tail (Fish, 1994). Fish (1994) hypothesized
sequence run, although this stroke may be that use of undulatory motions in conjunc-
modified to adjust for buoyancy (Costill et tion with abandonment of limbs and distal
al., 1992). Bipedal paddling is an adjust- expansion of the tail would increase swim-
ment of the quadrupedal gait by removal of ming performance. This would have cul-
fore- or hindlimbs movements during minated in the evolution of a high-efficien-
swimming. Semiaquatic mammals use bi- cy caudal hydrofoil with a lift-based oscil-
pedal paddling to reduce interference be- latory mode as displayed by cetaceans and
tween ipsolateral limbs and increase swim- sirenians. It is interesting that ancestral ce-
ming efficiency (Fish, 1 9 9 3 ~ ) .

FROM DRAG-BASED TO LIFT-BASED PROPULSION 635

-LH
0) LF
RF
n0

RH

-c1 LH

C
rm LF
IC-

-n RF

Q)
W RH

E LH
a
u) LF
u)
RF
n0
0 RH

I I I II II I I I I
20 40 60 80 I00
0

YO Cycle

FIG. 5. Gait diagrams for three terrestrial mammals swimming quadrupedally. Mammals include dog (Canis
@niliaris), elephant (Elephas maximus), and opossum (Didelphis virginiunu). Horizontal bars indicate the time
as a percent of cycle of power phase for the left hindfoot (LH), left forefoot (LF), right forefoot (RF), and right
hindfoot (RH). Opossum swam using a modified diagonal sequence run; elephant and dog swam with a running
lateral sequence run. Data from Fish (personal observation, 1993b).

taceans had long, robust tails (Gingerich et on land. Its ability to use all the flippers in
al., 1994; Thewissen et al., 1994). a walking gait are a consequence of the ori-
entation of the hindflippers. The hindflip-
In the absence of an elongate tail, lift- pers can be rotated under the body, similar
based propulsion is confined to oscillation to otariids seals. Phocid seals can not po-
of the limbs as in pinnipeds. Sea lions swim sition the hindflippers as in otariids and
by simultaneous motions of the foreflippers odobenids. Aquatic motions of phocids
(Feldkamp, 1987b) which may have devel- have no similarity with their terrestrial mo-
oped from a rowing posture using a modi- tions. The hindlimbs are not used in terres-
fied bound. Indeed at fast terrestrial speeds trial locomotion, but dragged passively
(2.6 d s e c ) , New Zealand fur seals (Arcto- (Tarasoff et al., 1972).
cephalus forsteri) use a bound (Beentjes,
1990). TRANSITION MODEL

Pelvic oscillations by both phocids and How did propulsive modes change in the
odobenids indicates use of similar motor evolution of the most derived aquatic mam-
patterns, although this may have been from mals? To deal with this question, a model
different evolutionary pathways. The rela- is proposed that demonstrates how transi-
tionship among pinnipeds is still controver- tion in swimming mode may have occurred
sial ( i e . , monophyletic versus polyphyletic; through the various intermediate forms
Barnes et al., 1985; Wyss, 1988, 1989). The (Fig. 6 ) . The problem of examining transi-
hindflippers alternate through power and re- tion from terrestrial to semiaquatic to fully
covery phases during the stroke cycle (Ray, aquatic is that rarely is there a complete set
1963; Tarasoff et al., 1972; Gordon, 1981; of extant intermediate forms to evaluate
Fish et al., 1988). This motion suggests a performance within the clade. Although a
motor pattern based on an asymmetrical number of important transitional fossils
gait. Odobenus uses a lateral sequence walk

636 FRANK E. FISH

PELVIC CAUDAL PECTORAL
OSCILLATION OSCILLATION
OSCILLATION
t
t DORSOVENTRAL
LATERAL
UNDULATION UNDULATION PECTORAL
ROWING
t f
ALTERNATE SIMULTANEOUS
PELVIC PELVIC

ROWING PADDLING /” SUBMERGED.........-............I....__....

‘ C - f................................ ........................... SURFACE
ALTERNATE ALTERNATE
PELVIC PECTORAL
PADDLING
PADDLING

QUADRUPEDAL
PADDLING

t

TERRESTRIAL
QUADRUPED

FIG. 6 . Model for sequence of transitions in swimming mode from drag-based paddling to lift-based oscillation.
Dashed line indicates change from surface to submerged swimming.

have been described recently (Gingerich et could be employed. Bipedal paddling pre-
al., 1983, 1990, 1994; Barnes et al., 1985; vents mechanical and hydrodynamic inter-
Berta et al., 1989; Berta and Ray, 1990; ference between the limbs and frees one set
Thewissen et al., 1994), information is lim- of limbs for locomotor stabilization, ma-
ited on how the shift from paddle propul- neuverability, tactile reception, and food
sion to lift-based propulsive modes may capture and processing.
have occurred. The model, therefore, draws
on the performance of modern species as With the development of semiaquatic
analogs of the primitive intermediate forms. habits using pectoral paddling or pelvic
Lauder (1995) justified this technique of us- paddling, there would be limb specializa-
ing modern analogs to determine perfor- tions for increased thrust production and
mance in extinct forms from a different clade. stability in water. These specializations
would include (1) development of short, ro-
The model assumes that an ancestral ter- bust humerus and femur, (2) elongation of
restrial mammal would first swim by pad- digits, ( 3 ) increased propulsive foot area by
dling quadrupedally using a terrestrial sym- addition of interdigital webbing or fringe
metrical gait (Rayner, 1985; Fish, 1993a). hairs, and (4) increased bone density for
This swimming mode would have allowed buoyancy control (Tarasoff et al., 1972;
the forelimbs to generate lift to keep the Stein, 1981, 1989; Wyss, 1988; Berta and
nares above the water surface for continued Ray, 1990; Fish and Stein, 1991; Thewissen
respiration. The addition of a non-wettable et al., 1994).
fur provides a positive buoyancy to semi-
aquatic mammals allowing them to float at Underwater foraging would necessitate a
the water surface. Non-wettable fur could change in mode to compensate for the pos-
have uncoupled buoyancy control for itive buoyancy from the animal and its fur
breathing from swimming movements (see section below on Swimming depth and
(Fish, 1993~)W. ithout the need to generate mode transition). By re-orienting the limbs
lift, a more effective bipedal paddling and moving them in a rowing motion, thrust
mode, using either forelimbs or hindlimbs, can be generated while simultaneously ex-
erting a downward force. Mizelle (1935)

FROM DRAG-BASED TO LIFT-BASED PROPULSION 637

noted that muskrats change from paddling cations of the limbs into winglike structures

with the hindfeet at the surface to rowing (Wyss, 1988) would have led to a change

when submerged. In rowing, the symmet- from drag-based to lift-based propulsion.

rical terrestrial gaits are modified to limb The pectoral oscillatory stroke of sea lions,
4 motions restricted to the horizontal plane. Zulophus,has a distinct paddling phase in-

Another mechanism to generate force to terposed between power and recovery

counteract buoyancy would be to increase phases (Feldkamp, 1987b).

thrust by diving with simultaneous strokes The sequence by which caudal oscillation

of the paired limbs. Once submerged to a arose goes from simultaneous pelvic pad-

sufficient depth compression of the air en- dling through undulation of the body and

trapped in fur would reduce the buoyancy tail. Such intermediate states were demon-

on the animal. Enhancement of this action strated from subsurface swimming in otters

would be facilitated by undulation of the by Williams (1989) and Fish (1994). Like-

tail as is seen in river otters (Tarasoff et al., wise fossil evidence from early whales in-

1972; Fish, 1994). dicates the importance of the combination

In the evolution to a highly aquatic life- of hindlimb paddling and spinal undulation

style, buoyancy control for diving could be (Gingerich et al., 1994; Thewissen et al.,

effected by abandoning fur for blubber. 1994). Spinal undulation evolved before de-

This change reduces maintenance cost for velopment of a tail fluke as indicated by the

grooming and allows the animal to more fossil Ambulocetus natans (Thewissen et

easily maintain neutral buoyancy at depth. al., 1994). Eventually by eliminating use of

The exception is the sea otter which spends inefficient paddling appendages, the undu-

a significant amount of time grooming its latory mode would dominate and improve

fur and maintains its buoyancy with en- performance. The addition of caudal flukes

larged lungs (Kooyman, 1989; Williams, would further enhance performance in the

1989). final transition to lift-based swimming. The

Selection for increased energy efficiency presence of a long robust tail is a precon-

and higher swimming speeds would have dition for the development of this mode.

facilitated the evolution of lift-based pro-
pulsion once semiaquatic mammals had SWIMMING DEPTH AND MODE TRANSITION

adapted the drag-based modes for sub- While performance in regard to efficien-

merged swimming (Fig. 6). Pelvic oscilla- cy, speed, and thrust production are impor-

tion would have originated from pelvic tant in the evolution of swimming modes,

rowing. These modes show similarities in swimming depth is a major factor in the

the alternating movements of the hindlimbs effectiveness of the modes. This is no in-

which each experience power and recovery cidental statement in that mammals as ob-

phases as the limbs move in a horizontal ligate air-breathers must return to the water

plane. It is anticipated that an intermediate surface. Swimming at the air-water inter-

form using lateral undulations of the body face comes with an additional cost in the

with rowing action from the hindlimbs drag (Lang and Daybell, 1963; Hertel,

would have occurred in the transitional se- 1966; Fish, 1982; Williams, 1989). Surface

quence. Although no modem analog of this swimming can augment drag by as much as

swimming mode exists, Berta et al., (1989) five times (Hertel, 1966). Increased drag at

stated that in an the early fossil pinniped, the surface, termed wave drag, results from

Enaliarctos mealsi, the lumbar vertebrae al- construction of waves with energy lost to

lowed swimming by lateral movements of the water as potential energy in lifting it

, the trunk along with limbed propulsion. vertically (Vogel, 1994).
Pectoral oscillation also would have had its Alternate pectoral and alternate pelvic

origin from rowing. The motion of the fore- paddling modes appear to work well at the

limbs in the power phase of the stroke surface, where orientation of the limbs

would not only be directed posteriorly but helps to maintain stability (Fish and Stein,

ventrally as well. Simultaneous strokes of 1991) and the propulsive forces from the

the forelimbs in conjunction with modifi- limbs are effectively applied without sur-

638 FRANK E. FISH

face interference. Simultaneous paddling wave drag and limitations of hull speed

and rowing are utilized by semiaquatic which could impede an animal’s forward

mammals when submerged. These modes progress. Alternatively by leaving the water

are associated with force production to altogether by porpoising, marine mammals

counteract buoyancy and increased speed. can maintain high speed with reduced en-

Otters paddle with an alternating stroke at ergy costs (Au and Weihs, 1980).

the water surface, but switch to simulta- CONCLUDING REMARKS
neous strokes with body and tail undula-

tions when moving quickly underwater Semiaquatic mammals occupy the inter-

(Williams, 1989; Fish, 1994). Oscillatory mediate position between aquatic and ter-

movements by derived aquatic mammals restrial environments. Compared to mam-

are more effective underwater. While sub- mals specialized for land or water, semi-

merged, these modes would be prevented aquatic mammals are in the evolutionarily

from displacing water vertically and losing more precarious position of being unable to

energy to wave drag. specialize in their locomotor performance

Speed, in addition to energy economy, is in either environment. However, it was

impacted negatively by swimming at the these intermediate animals that would give

water surface. Interaction of waves gener- rise to more derived aquatic mammals.

ated by the swimming animal sets a limit Swimming behaviors of modern analogs of

to maximum speed, referred to as “hull these primitive intermediates best represent

speed.” Hull speed was original used to the original transitional stage. Comparison

measure the performance of ships, but has of modern semiaquatic mammals with fully

been applied to animal systems (Prange and aquatic mammals indicates which potential

Schmidt-Nielsen, 1970; Aigeldinger and selection factors and mechanical constraints

Fish, 1995). Hull speed is directly related may have directed the evolutionary course

to hull or body length and is due to the of derived aquatic forms.

constructive interference of the diverging The secondary radiation of mammals

waves set up at the bow and stem. As speed into aquatic habitats was possible through a

increases, the wavelength of the system in- combination of behavioral, physiological,

creases until the wavelength of the bow and morphological adaptations. The inter-

wave matches body length. The animal be- action between the habits of mammalian

comes trapped in the wave trough limiting swimmers and the aqueous environment has

further increases in speed (Vogel, 1994). An produced different solutions for effective

animal would have to literally swim uphill aquatic locomotion by various morpholog-

to swim faster, which is energetically costly ical designs and propulsive modes. Evolu-

(Prange and Schmidt-Nielsen, 1970). tionary transition from terrestrial to fully

Mammals rarely exceed hull speed when aquatic habits was accompanied by in-

swimming at the surface (Aigeldinger and creased swimming performance by sub-

Fish, 1995). Competitive human swimmers merged swimming, streamlined body de-

reach hull speed at their maximum swim- signs, and lift-based propulsive modes,

ming speed (Kolmogorov and Duplishche- which fosters increased speed, efficiency,

va, 1992; Videler, 1993). and thrust production while reducing drag

Mammals that swim submerged are not and cost of transport.

subjected to the same speed limitation as It is expected that swimming modes as-

surface swimmers. If a dolphin like Tur- sociated with semiaquatic and fully aquatic

siops with a body length of 2.6 m (Fish, mammals originated from terrestrial gaits.

19936) were limited to hull speed, its pre- Primitive quadrupedal and bipedal paddling

dicted maximum speed would be only 2 represent modified symmetrical gaits,

d s e c . However, Tursiops has been report- whereas rapid asymmetrical gaits were

ed to swim up to 15 d s e c (Lockyer and combined with oscillating hydrofoils. The

Morris, 1987; Fish, 1993b). By swimming transitional sequence of swimming mode

at a depth greater than three body diameters with increasing aquatic habits corresponds

(Hertel, 1966), there is no effect due to directly to the continuum of terrestrial gaits

FROM DRAG-BASED TO LIFT-BASED PROPULSION 639

associated with increasing speed. The idea their effect on the energy cost of swimming. J.
that neuromotor patterns are conservative is Zool., London 21 1:177-192.
the foundation to the terrestrial-aquatic Berta, A. 1991. New Enaliarctos (pinnipedimorpha)
transition. This conservatism permits large from the Oligocene and Miocene of Oregon and
scale changes in swimming kinematics and the role of “Enaliarctids” in pinniped phylogeny.
performance with only slight modification Smithsonian Contrib. Paleobiol. 69: 1-33.
of the neuromotor pattern originating from Berta, A,, C. E. Ray, and A. R. Wyss. 1989. Skeleton
terrestrial locomotion. Grillner (1 996) noted of the oldest known pinniped Enaliarctos mealsi.
that “evolution rarely throws out a good de- Science 244:60-62.
sign but instead modifies and embellishes Berta, A. and C. E. Ray. 1990. Skeletal morphology
on what already exists.” and locomotor capabilities of the archaic pinniped
Enaliarctos mealsi. J. Vert. Paleont. 10:141-1517,
ACKNOWLEDGMENTS Blake, R. W. 1983. Fish locomotion. Cambridge Uni-
versity Press, Cambridge.
I wish to thank J. H. Long and G. V. Blickhan, R. and J. Y. Cheng. 1994. Energy storage
Lauder for inviting me to participate in this by elastic mechanisms in the tail of large swim-
symposium. I appreciate the comments mers-a re-evaluation. J. Theor. Biol. 168:315-
321.
made by R*v. Baudinette and B- D. ‘lark Bose, N. and J. Lien. 1989. Propulsion of a fin whale
(Balaenopteru physalus): Why the fin whale is a
on the draft of the manuscript. Discussions fast swimmer. Proc. Roy. SOC.London B 237:
with G . E. Goslow, D. A. Pabst, S. Peters, 175-200.
J. G. M. Thewissen, and T. M. Williams Bose, N., J. Lien, and J. Ahia. 1990. Measurements
were extremely helpful in development of of the bodies and flukes of several cetacean spe-
the manuscript. The Coca-Cola Co. is ac- cies. Proc. Roy. SOC.London B 242:163-173.
knowledged for donation of the elephant Costello, R. R. and G. C. Whittow. 1975. Oxygen cost
swimming video. Some of the work sum- of swimming in a trained California sea lion.
marized in this paper was funded by ONR Comp. Biochem. Physiol. 50:645-647.
grant NOOO14-95-1-1045. Costill, D. L., E. W. Maglischo, and A. B. Richardson.
1992. Swimming. Blackwell, Oxford.
REFERENCES Davis, R. W., T. M. Williams, and G. L. Kooyman.
1985. Swimming metabolism of yearling and
Adachi, J. and M. Hasegawa. 1995. Phylogeny of adult harbor seals Phoca vitulina. Physiol. Zool.
whales: Dependence of the influence on species 58590-596.
sampling. Mol. Biol. Evol. 12:177-179. Dial, K. P., G. E. Goslow, Jr., and E A. Jenkins, Jr.
1991. The functional anatomy of the shoulder in
Aigeldinger, T L. and E E. Fish. 1995. Hydroplaning the European starling (Sfurnus vulgaris). J.
by ducklings: Overcoming limitations to swim- Morph. 207:327-344.
ming at the water surface. J. Exp. Biol. 198:1567- DiPrampero, F! E., D. R. Pendergast, D. W. Wilson,
1574. and D. W. Rennie. 1974. Energetics of swimming
in man. J. Appl. Physiol. 37:l-5.
Alexander, R. McN. 1988. Why mammals gallop. English, A. W. 1976. Limb movements and locomotor
Am. Zool. 28:237-245. function in the California sea lion (Zalophus cal-
iforniums). J. Zool., Lond. 178:341-364.
Au, D. and D. Weihs. 1980. At high speeds dolphins Feldkamp, S. D. 1987a. Swimming in the California
save energy by leaping. Nature <London)-284: sea lion: Morphometrics, drag and energetics. J.
548-550. Exp. Biol. 131:117-135.
Feldkamp, S. D. 19876. Foreflipper propulsion in the
Backhouse, K. M. 1961. Locomotion of seals with California sea lion, Zalophus californianus. J.
particular reference to the forelimb. Symp. 2001. 2001. Lond. 212:43-57.
SOC.London 559-75. Fish, E E. 1982a. Function of the compressed tail of
surface swimming muskrats (Ondatra zibethicus).
Barnes, L. G., D. P. Domning, and C. E. Ray 1985. J. Mamm. 63591-597.
Status of studies on fossil marine mammals. Mar. Fish, E E. 19826. Aerobic energetics of surface swim-
Mamm. Sci. 1:15-53. ming in the muskrat Ondatra zibethicus. Physiol.
2001. 55:180-189.
Baudinette, R. V.and E! Gill. 1985. The energetics of Fish, E E. 1984. Mechanics, power output and effi-
ciency of the swimming muskrat (Ondatra zibeth-
‘flying’ and ‘paddling’ in water: Locomotion in icus). J. Exp. Biol. 110:183-201.
penguins and ducks. J. Comp. Physiol. 155:373- Fish, E E. 1985. Swimming dynamics of a small
380. semi-aquatic mammal. Am. 2001.25: 13A.
Beentjes, M. F! 1990. Comparative terrestrial loco- Fish, E E. 1992. Aquatic locomotion. In T. E. Tomasi
motion of the Hooker’s sea lion (Phocarctos hook- and T H. Horton (eds.), Mammalian energetics:
eri) and the New Zealande fur seal (Arctocephalus Interdisciplinary views of metabolism and repro-
forsteri): Evolutionary and ecological implica-
tions. Zool. J. Linn. SOC.98:307-325.
Bennett, M. B., R. E Ker, and R. McN. Alexander.
1987. Elastic properties of structures in the tails
of cetaceans (Phocaena and Lagenorhynchus) and

640 FRANK E. FISH

duction, pp. 34-63. Cornell Univ. Press, Ithaca, Kenyon, K. W. 1969. The sea otter in the eastern Pa-
New York. cific Ocean. N. Amer. Fauna 68: 1-352.
Fish, E E. 1 9 9 3 ~ .Influence of hydrodynamic design
and propulsive mode on mammalian swimming Kolmogorov, S. V. and 0. A. Duplishcheva. 1992.
energetics. Aust. J. Zool. 42:79-101. Active drag, useful mechanical power output and
Fish, E E. 1993b. Power output and propulsive effi- hydrodynamic force coefficient in different swim-
ciency of swimming bottlenose dolphins (Tursiops ming strokes at maximal velocity. J. Biomech. 25:
truncarus). J. Exp. Biol. 185:179-193.
Fish, E E. 1993c. Comparison of swimming kine- 31 1-318.
matics between terrestrial and semiaquatic opos- Kooyman, G. L. 1989. Diverse divers: Physiology
sums. J. Mamm. 74:275-284.
Fish, E E. 1994. Association of propulsive swimming and behavior. Springer-Verlag, Berlin.
mode with behavior in river otters (Lutra cana- Kruse, D. H. 1975. Swimming metabolism of Cali-
densis). J. Mamm. 75:989-997.
Fish, E E., S. Innes, and K. Ronald. 1988. Kinematics fornia sea lions, Zulophus californianus. M.S.
and estimated thrust production of swimming harp Thesis, San Diego State University.
and ringed seals. J. Exp. Biol. 137:157-173.
Fish, E E. and C. A. Hui. 1991. Dolphin swimming- Lang, T G. 1966. Hydrodynamic analysis of dolphin
a review. Mamm. Rev. 21:181-195. fin profiles. Nature 209: I 110-1 11I.
Fish, E E. andB. R. Stein. 1991. Functional correlates
of differences in bone density among terrestrial Lang, T. G. and D. A. Daybell. 1963. Porpoise per-
and aquatic genera in the family Mustelidae formance tests in a seawater tank. NOTs Techni-
(Mammalia). Zoomorphology 110:339-345. cal Publication 3063. Naval Ordnance Test Sta-
Gingerich, l? D., N. A. Wells, D. E. Russell, and S. M. tion, China Lake, CA. NAVWEPS Report 8060.
I. Shah. 1983. Origin of whales in epicontinental
remnant seas: New evidence from the early Eo- Lauder, G. V. 1990. Functional morphology and sys-
cene of Pakistan. Science 220:403-406. tematics: Studying functional patterns in a histor-
Gingerich, F! D., B. H. Smith, and E. L. Simons. 1990. ical context. Ann. Rev. Ecol. Syst. 21:317-340.
Hind limbs of Eocene Basilosuurus: Evidence of
feet in whales. Science 249:154-157. Lauder, G. V. 1995. On the inference of function from
structure. In J. J. Thomason (ed.), Functional mor-
Gingerich, P.D., S. M. Raza, M. Arif, M. Anwar, and phology in vertebrate paleontology, pp. 1-18.
X. Zhou. 1994. New whale from the Eocene of Cambridge University Press, Cambridge.

Pakistan and the origin of cetacean swimming. Lighthill, J. 1969. Hydrodynamics of aquatic animal
Science 368:844-847. propulsion-a survey. Ann. Rev. Fluid Mech. 1:
Gordon, K. R. 1981. Locomotor behaviour of the wal- 413-446.
IUS (Odobenus). J. Zool., Lond. 195:349-367.
Goslow, G. E., Jr., K. F! Dial, and E A. Jenkins, Jr. Lindsey, C. C. 1978. Form, function, and locomotory
1989. The avian shoulder: An experimental ap- habits in fish. In W. S. Hoar, and D. J. Randall
proach. Amer. Zool. 29:287-301. (eds.), Fish physiology: Locomotion, Vol. 7, pp.
Grillner, S. 1975. Locomotion in vertebrates: Central 1-100. Academic Press, New York.
mechanisms and reflex interactions. Physiol. Rev.
55:247-304. Lockyer, C. and R. Morris. 1987. Observations on
Grillner, S. 1996. Neural networks for vertebrate lo- diving behaviour and swimming speeds in wild
comotion. Sci. Amer. 274:64-69. juvenile Tursiops truncatus. Aqu. Mamm. 13.1:
Grillner, S. and €? Wallen. 1985. Central pattern gen- 31-35.
erators for locomotion, with special reference to
vertebrates. Ann. Rev. Neurosci. 8:233-261. Milinkovitch, M. C., G. Orti, and A. Meyer. 1993.
Hartman, D. S. 1979. Ecology and behavior of the Revised phylogeny of whales suggested by mito-
manatee (Trichechus manatus) in Florida. Spec. chondrial ribosomal DNA sequences. Nature
Pub. Amer. SOC.Mammal. 5:l-153. (London) 361:346-348.
Hertel, H. 1966. Structure,form,movement. Reinhold,
New York. Mizelle, J. D. 1935. Swimming of the muskrat. J.
Holmer, I. 1972. Oxygen uptake during swimming in Mamm. 16:22-25.
man. J. Appl. Physiol. 33502-509.
Howell, A. B. 1930. Aquatic mammals. Charles C. Nishiwaki, M. and H. Marsh. 1985. Dugong-Du-
Thomas, Springfield, Illinois. gong dugon (Muller, 1776). In S. H. Ridgway and
Innes, H. S. 1984. Swimming energetics, metabolic R. Hamson (eds.), Handbook of marine mammals,
rates and hind limb muscle anatomy of some pho- Vol. 3, pp. 1-31. Academic Press, London.
cid seals. Ph.D. Diss., University of Guelph, On-
tario. 0ntsland, N. A. and K. Ronald. 1975. Energetics of
Jenkins, E A,, Jr. and G. E. Goslow, Jr. 1983. The the free diving harp seal (Pagophilus groenlun-
functional anatomy of the shoulder of the savan- dicus). Rapp. P.-V. Reun. Cons. Int. Explor. Mer.
nah monitor lizard (Vuranus exanthematicus). J. 169:451-454.

Morph. 175:195-216. Pabst, D. A. 1996. Morphology of the subdermal con-
nective tissue sheath of dolphins: A new fibre-
wound, thin-walled, pressurized cylinder model
for swimming vertebrates. J. Zool. (London) 238:
35-52.

Pearson, K. G. 1993. Common principles of motor
control in vertebrates and invertebrates. Ann. Rev.
Neurosci. 16:265-297.

Peters, S. E. 1983. Postnatal development of gait be-
haviour and functional allometry in the domestic
cat (Felis catus). J. Zool., Lond. 199:461-486.

Prange, H. and K. Schmidt-Nielsen. 1970. The met-
abolic cost of swimming in ducks. J. Exp. Biol.
53:763-777.